AddBack

1. One-liner

Proves every add-back with a source document before the buyer’s QoE analyst strips it from your sale price.

2. Trend signal — why now?

The sell-side of small-business M&A got materially more adversarial in the last 18 months, and the seller’s paperwork didn’t keep up.

Three things changed:

Buyers now run QoE on deals that never got one. Quality-of-earnings reviews are “standard practice in M&A transactions above $2M EBITDA and increasingly common for $1M+ deals.” That’s a downward march into a wallet that was never QoE-exposed. And the tone shifted: QoE reviews in 2026 dental practice sales are described as “more confrontational than in prior years” and “frequently challenge EBITDA adjustments that were not questioned in 2022–2023.”

The rejection rate is brutal and quantified. Buyers’ QoE consultants reject 30–50% of add-backs. Sell-side QoE reports typically reduce a seller’s stated EBITDA by 10–30%. Add-back disputes are named as “the #1 reason LOIs re-trade.”

Preparation timing is the single controllable variable, and it’s measured. Owners who start preparing 12–18 months before going to market “routinely retain 80 to 90 percent of their initial addbacks in QoE.” Owners who assemble the schedule the week of LOI retain “50 to 70 percent.” That delta is not a rounding error — at a 3.0x SDE multiple, every $50K of rejected add-back is $150,000 of purchase price, and the gap “routinely runs $200,000 to $800,000 on larger transactions.”

The mechanism is boring and specific. A QoE analyst ties each add-back to source documents — payroll registers, lease agreements, mileage logs, engagement letters, credit card statements. The rule of thumb quoted in the field is “no log, no addback.” A legitimate add-back with no paper “becomes a negotiating problem regardless of how valid it is,” because “buyers will not accept an add-back on your representation alone. They need proof.”

And the seller almost always meets that request list too late: “Most sellers encounter the due diligence checklist for the first time after signing a letter of intent, when the buyer’s advisor sends a request list and a deadline.”

Meanwhile the money confirms the pain is priced. Sell-side QoE runs $15,000–$25,000 for sub-$3M EBITDA and $25,000–$50,000 for $3M–$10M EBITDA — and it is sold on exactly this outcome, reportedly preventing “80%+ of buyer-side add-back disputes.” That’s a $15K floor on a service whose core job, for a small seller, is proving twenty line items. The seller with $158,950 of median cash flow cannot buy it.

Provenance:

• Signal 1 (demand): Buyers’ QoE consultants reject 30–50% of add-backs; add-back disputes are the #1 reason LOIs re-trade; $200k of disputed add-backs costs the seller $1.3M at a 6.5x multiple — https://ctacquisitions.com/adjusted-ebitda-add-backs-business-sale-2026/ — 2026-08-27
• Signal 2 (feasibility/behaviour): Owners preparing 12–18 months ahead retain 80–90% of add-backs vs 50–70% for week-of-LOI preparation; QoE analysts tie each add-back to source documents — payroll registers, engagement letters, mileage logs, credit card statements — https://ctacquisitions.com/sde-addbacks-explained-for-small-business-sellers/ — 2026-08-27
• Signal 3 (economic): Sell-side QoE costs $15,000–$25,000 for sub-$3M EBITDA and $25,000–$50,000 for $3M–$10M EBITDA, sold on preventing 80%+ of buyer-side add-back disputes; valuation gap is the #1 reason engagements end without a deal at 26% — https://www.bedrockqoe.com/insights/quality-of-earnings-report-cost + https://ctacquisitions.com/guides/small-business-ma-statistics-2026/ — 2026-08-27 Category: Underserved niche (a substantiation layer that exists only inside a $15K–$50K QoE engagement, with no self-serve tier for the seller below that threshold) + Workflow automation (per-line-item document matching in a market where the current tool is a spreadsheet and a shoebox)

3. The opportunity

Every tool in this market sells the number. Nobody sells the proof behind the number.

Look at the field. BizEquity gives you an instant valuation off an algorithm trained on a million valuations. Capitaliz walks you through a 21-step planning process with a Dynamic Revaluation™ engine. The Value Builder System scores you across 8 value drivers and hands the advisor a report. ExitAdviser sells a $49–$299/mo FSBO workflow with checklists and benchmarks. All of them are, structurally, scoring and assessment products. They will happily tell you your adjusted EBITDA is $1.2M.

None of them will tell you that the $38,000 of “consulting fees” inside that number has no engagement letter behind it, and will be gone in week three of diligence.

That’s the gap. The valuation is an output. The add-back schedule is an assertion. And the assertion is what gets attacked. The seller’s actual failure mode isn’t “I don’t know what my business is worth” — it’s “I claimed $300,000 of adjustments and the buyer validated $150,000,” which the field describes plainly: normalized EBITDA drops from $1.2M to $1.05M and “at a 4x multiple, that is $600,000 in reduction of your sale price.”

The incumbent that does solve this is the sell-side QoE firm — and it solves it well, at $15K–$50K, in 4–8 weeks, for a seller with $1M+ of EBITDA. That’s a real service and I’m not trying to beat it on quality. I’m pointing at everyone underneath it. The median BizBuySell-tracked business sells at $350,000 with $158,950 of cash flow. A $15,000 QoE on that deal is 4% of enterprise value to defend a schedule with maybe fifteen line items on it. Nobody buys that, so nobody does it, so those sellers walk into diligence with a spreadsheet and lose 30–50% of their add-backs on documentation grounds alone.

The 10× move is unglamorous: take the seller’s existing accounting file and card statements, find every transaction that plausibly supports each claimed add-back, attach it, flag the ones with nothing behind them, and produce a docket a QoE analyst can tick through. That is document retrieval and classification against a known taxonomy — the categories are public and stable (owner comp, owner perks, one-time professional fees, related-party rent above market, non-cash items) as are the rejects (recurring “one-time” costs, vague consulting fees, aspirational marketing, family at market rate, forward-looking run-rate adjustments).

This is the priced-out, not unserved shape. The service exists. It has a $15,000 floor. The population below the floor is large, feels the exact same pain, and loses six figures to it.

4. Target market

Primary customer: The US owner of a $700K–$8M revenue business — home services, HVAC, mechanical, landscaping, small manufacturing, professional practice, e-commerce — who is 6–24 months from listing, with $150K–$1.5M of SDE/EBITDA and an add-back schedule of 10–40 line items. Typically the owner and a part-time bookkeeper. No CFO. Books in QuickBooks, personal and business spending genuinely commingled because that’s how a closely-held company actually runs.

Secondary and probably the better commercial motion: the business broker or M&A advisor who lists 8–30 of these a year, already performs financial recasting as part of their commission, and eats the retrade when the schedule doesn’t hold.

Why they buy, in the market’s own words: “Buyers will not accept an add-back on your representation alone. They need proof!” And: deals “die in due diligence, often because the seller could not produce a document fast enough or produced two versions of the same number.” Sellers themselves describe diligence as “the business equivalent of a colonoscopy.”

Rough TAM reasoning: 9,586 small businesses closed through BizBuySell-tracked brokers in 2025 — but that “captures only the listed, broker-visible slice; total US transfers including off-market and family transitions are larger and not tracked by any single public source.” Axial alone had 12,856 businesses brought to market in 2025, up 17.1% YoY. Critically, only “20 to 30 percent of businesses that go to market ever sell” — so the population that attempts a sale in a given year is roughly 3–5× the closed-deal count, and every one of them builds an add-back schedule. Add the brokers: 90%+ of business brokers use BizBuySell, and each runs a book of listings. I’ll size the reachable annual buyer population in the low tens of thousands, not hundreds of thousands. This is a small market. It’s also a market where a single customer’s problem is worth six figures to them.

Why now for them: QoE marched down-market into the $1M-EBITDA band; scrutiny of 2022–2023-era adjustments got sharper; and the 12–18-month preparation window is now a published, quantified lever (80–90% vs 50–70% retention) that brokers are actively telling their sellers about. The advice exists everywhere. The tooling to execute it doesn’t.

5. Product sketch (MVP)

• Add-back schedule builder — the seller lists claimed adjustments in plain language (“my truck,” “wife’s payroll,” “the lawsuit,” “country club”), and each becomes a tracked line item with a category and a dollar amount.
• Evidence matcher — connects the QuickBooks file and card/bank statements, and for each claimed add-back proposes the specific transactions, payroll entries, and invoices that support it. The seller confirms or rejects the match.
• Substantiation status per line — every add-back carries a plain traffic light: documented (source attached), thin (transactions found, no invoice or agreement), undocumented (nothing behind it). This is the whole product in one screen.
• Kill-list flagging — automatically flags line items in the categories buyers consistently reject: expenses tagged “one-time” that recur annually, consulting fees with no engagement letter, owner travel with a business component, family payroll at market rate, forward-looking run-rate adjustments.
• Missing-document chase list — the exact list of papers to go find, ranked by dollars at risk. “Get the engagement letter for the $38K legal matter — $114,000 of purchase price at 3.0x.”
• Benchmark helper for comp and rent — for the two adjustments that require market comparison (excess owner compensation, related-party rent above market), attach a third-party reference point rather than an assertion.
• The docket export — a buyer-facing PDF: each add-back, its amount, its narrative, its category, and its attached source documents, in the order a QoE analyst works. Plus a private seller version showing what’s still exposed.
• Broker workspace — an advisor sees every seller in their book with a single number: percentage of claimed add-back dollars currently substantiated.

6. AI angle — what’s load-bearing

Remove the AI and this is a folder with subfolders. Which is what sellers already have, and it’s why they lose the money.

The load-bearing work is matching a vague human claim to specific financial records. The seller says “my truck.” That has to become: this lease payment line recurring monthly in the P&L, these fuel charges on the business card, this insurance premium — and then the flag that there is no mileage log, which is fatal, because “no log, no addback.” The seller says “the lawsuit was one-time.” The model has to go look at three prior years and notice legal fees appear every single year, which is precisely the pattern buyers reject as a recurring “one-time” expense.

That’s three genuinely model-shaped jobs: reading heterogeneous documents (statements, invoices, engagement letters, payroll registers) into structured line items; matching a natural-language claim to a set of transactions across accounts with no shared key; and pattern-testing a “non-recurring” assertion against multi-year history. A rules engine can’t do the first or second — the categories are stable but the documents are chaos, and every small business codes its chart of accounts differently.

What AI is explicitly not doing here: deciding whether an add-back is legitimate. It proposes evidence and flags exposure. The seller and their advisor decide. That boundary matters legally and it matters for trust.

7. Localization angle (if any)

N/A — this is a US play. The add-back taxonomy, the SDE-vs-EBITDA convention, the QoE norm, the SBA-financed buyer, and the broker infrastructure are all US-specific. The UK and Australia have analogous lower-middle-market M&A with different adjustment conventions and would need a separate product, not a translation. Deliberately not chasing that.

8. Business model — path to $1M–$5M ARR

Pricing:

• Seller, self-serve: $149/mo while preparing, minimum 3 months. Most sellers run 6–18 months. Call it ~$1,200 average lifetime.
• Seller, done-with-you: $2,500 one-time — the docket assembled and reviewed. This is the volume seller of the two, because the buyer is a 58-year-old HVAC owner who does not want a subscription, wants the artifact, and is comparing against a $15,000 QoE.
• Broker/advisor seat: $399/mo per advisor for unlimited seller workspaces in their book, plus white-label on the docket export.

ACV: ~$2,400 blended for sellers (mix of subscription and one-time), $4,800 for a broker seat.

Rough math to $1M ARR: 120 broker seats ($576K) + ~180 done-with-you dockets a year ($450K) = $1.03M. That’s 120 advisors out of a US broker population in the thousands, and 180 dockets out of a population that lists tens of thousands of businesses a year. Both numbers are small fractions of the addressable base.

Rough math to $5M ARR: needs the broker channel to become the default — roughly 400 advisor seats and 1,200 dockets a year — plus an attached second product. The obvious one is the full diligence document room (the request list arrives after LOI and the seller is unprepared for all of it, not just add-backs), which raises ACV without new distribution.

Expansion path: seller → broker seat → broker’s whole book → the post-LOI diligence request list. The natural ceiling is real: this is a one-transaction-per-customer product on the seller side, so retention comes from the advisor, not the owner. I’ve scored revenue and defensibility accordingly.

The pricing argument writes itself. $2,500 against a documented $200,000–$800,000 swing, benchmarked against a $15,000–$50,000 QoE that the customer has already been quoted and declined. This is the easiest value conversation in the portfolio.

9. Go-to-market wedge — first 100 customers

The customer is not hiding. They’re on a public list.

1. Work the broker directories, not the sellers. BizBuySell publishes a searchable business broker directory by city and 90%+ of brokers use it. Pull 2,000 US brokers, filter to those with 5+ active listings. Send each one a personalized artifact, not a pitch: take one of their live public listings, and send a one-page teardown of the add-back categories a QoE analyst would attack on a business of that type and size. That’s a two-minute AI-assisted job per broker and it demonstrates the entire product. Brokers already do recasting and already eat retrades; 3–5% booking a call on a list of 2,000 is 60–100 conversations. Close 20% into a $399/mo seat and that’s the first 12–20 broker customers, each carrying 8–30 sellers.

2. Intercept the sellers who already got the QoE quote and flinched. The exit-planning and QoE content ecosystem is enormous and highly commercial — CT Acquisitions, Bedrock QoE, dozens of broker blogs all publish 2026 pricing guides. Buy the long-tail search around the rejection, not the product: “why did the buyer reject my add-backs,” “add-back documentation checklist,” “quality of earnings cost small business.” These are people mid-transaction with six figures on the line. A free “add-back exposure scan” — upload the schedule, get the kill-list flags free, pay for the substantiation — converts on urgency, not education.

3. The Exit Planning Institute channel. CEPA-certified advisors are an organized, conference-attending, tooling-hungry population — the EPI itself publishes roundups of software tools for exit planning practices. Sponsor one event, get into one of those roundups, and land 10–15 advisor seats from a community that resells to owners for a living.

4. Partner with the fractional CFO and bookkeeping firms serving $1–10M businesses. They’re already in the QuickBooks file, they get asked “help me get ready to sell,” and they have no product to hand over. Revenue-share the done-with-you docket. This is a referral motion with maybe 30 firms, not 3,000.

5. The retrade post-mortem as content. Every broker has a war story about a deal that lost $400K in diligence. Collect ten of them with permission, publish them with the specific line item that died and the document that would have saved it. That is the highest-intent content in this market and nobody is writing it, because the people who know are selling $15K QoE reports to bigger clients.

10. Build complexity — justification

Low. The hard parts are all off-the-shelf in 2026: QuickBooks Online has a mature API, bank and card statement ingestion is a solved commodity, and document extraction from invoices, engagement letters, and payroll registers is exactly what current models do well. The add-back taxonomy is public, stable, and small — roughly a dozen accepted categories and eight rejected ones, all documented in the trade literature.

There’s no novel infrastructure. The genuinely fiddly work is the matching UX: making it fast for a non-financial owner to confirm “yes, those seven charges are the truck” across a few hundred candidate transactions, and making the exposure screen legible enough that the seller acts on it. That’s design iteration, not engineering risk.

Two engineers, 10–12 weeks to a docket a real broker will put in front of a real buyer. The domain knowledge is the scarce input, not the code — which is why I’ve tagged this domain-expertise-required and why the first hire or advisor should be someone who has actually sat on the sell side of twenty of these.

11. Gating checklist

Gate	Pass?	Note
Legal in target market	✅	Document organization and evidence matching. Not an attest service, not an audit, not a valuation opinion. Must be marketed carefully — see risk flags.
Ethical — no harm / dark patterns	✅	The product’s core act is flagging unsupported claims and telling the seller to drop or document them. It reduces the number of unsupportable assertions put in front of a buyer, it doesn’t manufacture them.
Market exists (evidence above)	✅	$15K–$50K incumbent service sold on this exact outcome; 30–50% rejection rate; 26% of failed engagements die on the valuation gap.
1–5 person team can build this	✅	Two engineers plus a domain advisor, 10–12 weeks.
Launchable with <$50K / ₹40L	✅	Standard SaaS build. No data acquisition cost, no regulatory spend, no inventory.

All five pass.

12. Feasibility score

Axis	Weight	Score	Notes
Problem intensity	20	17/20	Six-figure, quantified, and felt at a specific moment. $200K of disputed add-backs = $1.3M at 6.5x. Not a 20: it’s episodic, not daily — the owner feels it once per lifetime, which caps urgency outside the transaction window.
Demand evidence	15	12/15	Strong: a $15K–$50K incumbent service sold on precisely this outcome, published rejection rates, “#1 reason LOIs re-trade.” Held below 13 because the evidence is broker/advisor trade literature — commercially motivated sources — rather than raw seller voice. I could not source verbatim seller complaints; that’s a real gap.
Build feasibility	15	13/15	QuickBooks API, statement ingestion, document extraction all commodity. Matching UX is the only genuine design risk. 10–12 weeks.
Distribution clarity	15	11/15	The broker directory is a named, enumerable list with a personalized artifact that costs two minutes to produce. Not higher because broker adoption of new tooling is historically slow and the seller-side motion is one-shot, non-compounding.
Revenue mechanics	15	11/15	Pricing is benchmarked directly against a known $15K alternative — the easiest value argument in the portfolio. Capped at 11 because seller-side revenue is inherently one-transaction and $1M ARR leans hard on the broker seat holding.
Time to first revenue	10	6/10	Done-with-you dockets can be sold manually inside 8 weeks. But the seller is mid-transaction on their own timetable and brokers buy on a quarterly cadence — this is a 2–3 month motion to real revenue, not 4 weeks.
Defensibility	10	4/10	Honest score. The taxonomy is public, the workflow is copyable, and a QoE firm could ship a self-serve tier. What accrues is a corpus of which add-backs actually survived diligence by category and deal size — genuinely valuable at month 24, worthless at month 3. Execution-only moat for the first year.
Total	100	74/100

13. Qualitative modifiers

Founder-fit tags

technical-heavy · domain-expertise-required

The build is standard. The product is not buildable well by someone who hasn’t watched a QoE analyst dismantle a schedule. Get a sell-side advisor or a transaction-services CPA as a co-founder or first advisor before writing the taxonomy.

Key assumptions to validate (3–5)

1. Assumption: Brokers will pay $399/mo for something they currently absorb inside their commission. How to test: Send the personalized listing teardown to 100 brokers pulled from the BizBuySell directory. Measure reply rate and, specifically, how many say “we already handle this internally.” If >60% say that and won’t take a call, the broker channel is a mirage and the whole revenue model has to move to the seller.
2. Assumption: A seller will hand over their QuickBooks file and personal card statements to a stranger’s product. How to test: Run 15 done-with-you dockets manually, at $2,500, before building anything. Count how many stall at the data-sharing step. This is the highest-risk assumption in the idea and it’s cheap to test.
3. Assumption: The evidence matcher is accurate enough that confirming matches is faster than doing it by hand. How to test: Take 10 real add-back schedules with their underlying files, run the matcher, and time a non-financial person confirming the output vs. assembling from scratch. Needs to be 5× faster to justify the product’s existence.
4. Assumption: The docket actually moves the outcome — buyers accept more add-backs when they arrive documented. How to test: Track 20 dockets through to close and compare claimed-vs-accepted add-back dollars against the published 50–70% unprepared baseline. This is the only proof that matters and it takes 6–12 months to get.

Risk flags

1. Market size ceiling. This is the flag I’d worry about most. 9,586 broker-tracked closes in 2025; Q2 2026 transactions were down 10% both QoQ and YoY. The unlisted and off-market population is larger but unmeasurable, so I’m sizing on faith at the edges. A $5M ARR outcome requires near-total broker-channel penetration or a second product. $1–2M is the realistic honest target.
2. Regulatory/positioning risk. Get anywhere near “we validate your earnings” and you’re implying an attest function that requires a CPA license. The product must be positioned unambiguously as document organization and exposure flagging, with the seller and their advisor making every judgment call. Sloppy marketing here is a genuine legal problem, not a branding one.
3. Incumbent counterattack. A QoE firm with existing brand and referral flow can launch a $2,500 light-touch tier and use the full $25K report as the upsell. They’d have distribution I’d have to build. The defense is speed and owning the sub-$1M-EBITDA band they don’t want.
4. One-shot customer. The seller churns by definition — they sell the business and leave. Every dollar of durable ARR lives in the broker seat. If assumption #1 fails, this is a services business with software leverage, not a SaaS company. That’s not fatal, but it’s a different business than the one scored here.
5. Evidence quality is commercially motivated. Nearly every statistic above comes from firms selling QoE reports, brokerage services, or exit planning. The 30–50% rejection figure and the 80–90%-vs-50–70% retention split are consistent across independent sources, which is reassuring, but all of those sources profit from sellers believing them. Discount accordingly — this is why confidence is Medium, not High.

14. Structured verdict

Score:                  74/100
Verdict:                GO
Confidence:             Medium
Best-fit builder:       Technical founder paired with a sell-side M&A advisor or
                        transaction-services CPA who has personally worked QoE
                        engagements on sub-$10M deals
Time to revenue:        8–12 weeks (manual done-with-you dockets before build)
Capital to launch:      $15–25K
Top 3 assumptions to validate first:
  1. Brokers pay for what they currently absorb — 100 personalized listing
     teardowns from the BizBuySell directory, measure reply and "we do this
     already" rate
  2. Sellers share the QuickBooks file and personal card statements — 15 manual
     $2,500 dockets, count stalls at the data-sharing step
  3. Matching beats manual by 5× — 10 real schedules, timed head-to-head against
     a non-financial person assembling from scratch
Kill criteria:
  - Abandon if <5 of 100 directory-sourced brokers take a call, or if >60% say
    they handle recasting internally and won't pay
  - Abandon if fewer than 8 of 15 manual docket buyers complete the data handover
  - Abandon if a QoE firm or Capitaliz/Value Builder ships a per-line-item
    substantiation tier under $3K before v1 ships
  - Abandon if tracked dockets show no improvement over the published 50–70%
    unprepared add-back retention baseline after 20 closed deals

15. Next step — 1-week validation sprint

• Day 1–2: Pull 300 brokers from the BizBuySell directory with 5+ active listings. For 100 of them, generate a one-page add-back exposure teardown against one of their live public listings — business type, size, the specific adjustment categories a QoE analyst would attack. Send all 100.
• Day 3–4: Call every broker who replies. One question drives everything: “When a deal retrades on add-backs, who eats it and what did you do about it last time?” Separately, get three of them to introduce me to a seller currently 6–12 months from listing.
• Day 5: Manually assemble one real docket for one real seller, free, in a day. Watch exactly where they stall on producing documents, and time the whole thing.

Falsifiable outcome: ≥5 of 100 brokers take a call AND ≥2 say they’d pay for a per-seller tool AND the manual docket surfaces at least 3 add-backs the seller could not document. Miss any of the three and the broker channel is wrong, which means the revenue model is wrong, and I go back to the shelf rather than trying to make the seller-only motion carry a SaaS business it can’t carry.